the use case is simple:
Customer and agent having conversation over Whatsapp or SMS and it might be a lengthy conversation (happening over several hours), for instance customer asked an update on a particular issue and agent replied in 2 hours and then customer replies “Thank you” in another 5 hours. So we somehow should understand that this isn’t a message we should be delivering to agent as a new interaction. Or should we ? And if not, how it should be handled properly ?
My guess would be that scenario BOT somehow should understand that this is a tail message to the previous conversation and just handle it inside the scenario.
In case of email it’s easy, those are linked together by the code in subject
in case of Messengers, there is no session on the web site which GIID can be linked to
So technically each time after the agent closed the interaction the session ends and a new GIID is created, but the “Thank you” message which comes after should be treated as a part of the original interaction somehow.
Was a bot involved earlier in the conversation before the interaction made it to an agent?
If so, then you could evaluate the timestamps within the bot response RAW and calculate the difference between the scenario current timestamp. Use this value in combination with a “thank you” intent in an IF block before routing to an agent. If true, route it back to the bot to say something like “You’re welcome. If that’s all you need you can end the conversation by closing the chat window, or tell me what else I can help you with.”
If no bot was involved. Use an IF object at the top of the scenario to evaluate for phrases like “thank you”.
The chat configuration in tenant admin should be configured to end a chat interaction after a period of no response from the customer. 5 hours is way too much time. I have a scenario and config I can show you that uses the same chat scenario for both FB messenger and webchat. After a period of time, like 10 minutes, the system automatically tells the custmer “we haven’t heard from you in awhile…” then it disconnects.
In the case of SMS, you can use similar logic to the above. If there is a bot, just use a generic catch all intent for “Thank you” and respond with “You’re welcome. If that’s all you need you can just reply “i’m done”, or tell me what else I can help you with.” This approach is pretty easy and would operate universally for active, new, or old interactions.
I suppose it all depends on how the logic looks today. Feel free to reach out if you think I can be of assistance.
Hey Derek, the solution you suggested make sense, but i’m looking for something more generic, so we can in someway pre-build it for the customers.
I’m thinking of some sort of vocabulary (customizable) which could be checked before the conversation starts and if hits the vocabulary (full match or indirect hit) just added to the last conversation and do not start a new one.
I wonder how much that is a problem for the customers at all today. Maybe i’m exaggerating the issue…
thoughts ?
Apologies for my delayed response.
I guess I don’t see how this would be a Bright Pattern issue. The issue is with the underlying bot platform and is something that the bots would/could handle with very simple logic.